Optimal Route Driving for Leader-Follower Using Dynamic Particle Swarm Optimization

Abstract

The mobile robots rely on trajectory generation problem when they are navigating in several environments, for achieving the best path. One of the solution by using a heuristic method, named Particle Swarm Optimization (PSO). In the previous study, by using such method, the mobile robot can find the best route towards the target without collision, moreover, its simplicity in algorithms, implement easily and has few parameters to regulate. However, the PSO original algorithm can't guarantee to produce an optimal solution. Local optimum still occurs especially in complex and dynamic environments, due to premature convergence. It causes the mobile robot collisions with obstacles and generates the long path to the target. In this paper, dynamic PSO is developed by using dynamic inertia function in setting parameter to accelerate convergence and re-initialization of particles performed to overcome the premature convergence. The comparison with three algorithms, such as OPSO, GPSO, and DPSO have analyzed in this paper. The proposed DPSO algorithm produce the optimum solution faster with the convergence of fewer than 150 iterations in static obstacles and 200 iterations on the moving obstacle, 4% shorter traveled lengths, 13% more smooth, with fast processing and it guaranteed to avoid collisions and stable movement to achieve the target.